Multi-DSP System For Vehicle Wheelset Wear Detection Image Processing

Rail vehicle safety issues become increasingly prominent with the rapid development of railtransit. To ensure the safety of vehicles traveling, detecting vehicles go round on the wearing of theline components is one of the important measures. Increasing wheel set wear detection efficiency isa key technology in the field to be solved, and the image processing efficiency is a key factor in theimage-based wear detection system. It is significant to study the multi-channel DSP rounds on theimage processing solutions to improve the detection efficiency of wheel set wear in this paper.The main contents of the paper are:(1) Technical indicators analysis and system designIn this paper, real-time detection must achieve wheel vehicle under the speed of80km/h, Aftera detailed analysis of the technical indicators the program based on FPGA+DSP multi-channelwheel image processing is proposed. wheel set wear detection system consists of image acquisitionunit and processing unit, this paper focuses on an image processing unit, and image processing uniton wheels is researched and designed.(2) Multi-channel wheel image processing hardware platform designIn this paper, the hardware platform of multi-channel DSP wheel image processing is designed,The image processing unit consists of image data transmission, the organization module and imagedata processing module. a thorough study and detailed design on the high-speed data exchangeinterface between two modules, and a detailed planning of the DSP memory space are done in thispaper; at last, a image processing hardware system with the core of TMS320DM642DSP iscompletely developed and product.(3) Multi-channel wheel image processing software system implementationMulti-channel DSP wheel image processing system software design, the initial configurationunder DM642DSP/BIOS image processing software systems and multi-task thread design iscompleted, and made a detailed design on implementation of each task thread; Meanwhile, a briefanalysis on the wheel wear detection system for image processing algorithms, and the redesign ofthe preparation, packaging and optimization algorithms under DSP platform is completed.(4) Experimental Design and AnalysisExperimental test on the key parts for the designed hardware circuit system. a commissioningof wheel image processing algorithms on the entire system after the data interface simulation debug,The experimental results show that the designed system to complete the wheel image processing to meet the needs of the real-time detection when wheel vehicle under the speed of80km/h.